Safevoice: A Privacy-Preseving Multimodal Anonymous Crime Reporting System With Ai-Assisted Prioritization

Abstract

Crime reporting is an essential component of public safety; however, a significant number of incidents remain unreported due to fear of identity disclosure, lack of accessibility, and unreliable reporting mechanisms. Traditional crime reporting systems often require personal identification and continuous internet connectivity, making them ineffective in emergency or low connectivity scenarios. To address these challenges, this paper presents SafeVoice, a privacy preserving multimodal anonymous crime reporting system with AI assisted prioritization.

The proposed system enables users to submit reports anonymously using text, image, and audio inputs through a mobile application. Offline data storage and automatic synchronization ensure reliable operation in constrained network environments. User privacy is protected through cryptographic pseudonym generation, while reported data is securely stored in a cloud backend. A hybrid prioritization mechanism combining keyword-based rules and a baseline machine learning model assists authorities in identifying urgent cases. A web based administrative dashboard supports human verification and ethical decision making. Experimental results from pilot evaluation demonstrate improved reliability, accessibility, and usability of the proposed system.